Milling of zirconia nanoparticles in a stirred media mill

Article abstract of DOI:10.1111/j.1551-2916.2008.02579.x

Dispersions of three different types of zirconia nanoparticles were treated in a stirred media mill. The deployed surface modifier was present during milling and it established separating mechanisms between the particles. The combination of mechanical deagglomeration and chemical surface modification results in stable zirconia colloids with average particle sizes down to 9 nm. In addition to deagglomeration, the milling treatment also causes comminution of nanoparticles. This was indicated for the two coarser types of the examined particles, by increasing surface areas and decreasing primary crystallite sizes. Transmission electron microscopy of the particles confirmed the creation of smaller primary crystallites and a minority of small fragments, but the majority of particles do not undergo comminution into halves or fragments with similar size. Changes of the particles' phase composition, wear of milling media, amorphization of the particles to a small extent, and leaching of Y ₂O₃ dopant have been observed as side effects in the process and are characterized quantitatively. This work describes a process for nanoparticle deagglomeration and preparation of high quality colloids, and informs about occurring side effects, including approaches for their minimization.

Full text of DOI:10.1111/j.1551-2916.2008.02579.x

J. Am. Ceram. Soc., 91 [9] 2836–2843 (2008)
DOI: 10.1111/j.1551-2916.2008.02579.x
r 2008 The American Ceramic Society
ournal
J
Milling of Zirconia Nanoparticles in a Stirred Media Mill
Jens Adam,* Robert Drumm, Gabi Klein, and Michael Veithw
INM – Leibniz Institute for New Materials, 66123 Saarbruecken, Germany
Dispersions of three different types of zirconia nanoparticles
were treated in a stirred media mill. The deployed surface
modifier was present during milling and it established separat-
ing mechanisms between the particles. The combination of
mechanical deagglomeration and chemical surface modification
results in stable zirconia colloids with average particle sizes
down to 9 nm. In addition to deagglomeration, the milling treat-
ment also causes comminution of nanoparticles. This was
indicated for the two coarser types of the examined particles,
by increasing surface areas and decreasing primary crystallite
sizes. Transmission electron microscopy of the particles con-
firmed the creation of smaller primary crystallites and a minor-
ity of small fragments, but the majority of particles do not
undergo comminution into halves or fragments with similar size.
Changes of the particles’ phase composition, wear of milling
media, amorphization of the particles to a small extent, and
leaching of Y2O3 dopant have been observed as side effects in
the process and are characterized quantitatively. This work de-
scribes a process for nanoparticle deagglomeration and prepa-
ration of high quality colloids, and informs about occurring side
effects, including approaches for their minimization.
Cruz et al.,6 which balances attracting Van der Waals forces and
repulsing electrostatic forces. For the resulting electrostatic
stabilization, it can be concluded that repulsing forces are
optimized with increased surface charges and with a minimized
ionic concentration in the dispersion medium. The electrostatic
mechanism is disadvantageous for long-term stabilization7 and
at high solid contents of the dispersions. These restrictions can
be overcome by the steric stabilization, where adsorbed mole-
cules form a layer on the particles which cannot interpenetrate
completely.8 But, contrarily to electrostatic repulsion, this sep-
aration mechanism takes place at a very short distance, and as a
result, the sedimentation stabilization of low solid content sus-
pensions is limited. Combined electrosteric stabilization is the
third choice.7–10 For nanoparticles, with their large specific sur-
faces and strong agglomeration tendencies, electrosteric stabili-
zation is the only satisfying option in most of the cases.
Surface modifiers are used to provide a surface chemistry on
particles, which is suitable for colloidal stabilization and which
is needed for further processing.2 To name a few possible func-
tions, surface modification can be used for zeta-potential tailor-
ing2 (pH of the isoelectric point, sign, and amplitude of surface
charge), to change the hydrophilic–hydrophobic character, or to
provide suitable groupings for the chemical crosslinking of par-
ticles with a polymer network. Therefore, the surface modifier
must be chemically bound to the particle surface. For most of
the nanoparticle applications it is necessary to use small modifier
molecules (in comparison with polymer or polyelectrolyte dis-
persants) to saturate the large surface areas and not to insert a
large fraction of an additional substance into the processes and
material compositions.
3,6,9-Trioxadecanoic acid (TODA) is a suitable surface mod-
ifier for the electrosteric colloidal stabilization of zirconia nano-
particles in water, ethanol, and other dispersion media.3,8,9,11–13
The molecule is relatively small (178 g/mol) and can be chem-
ically bound to zirconia surfaces.3,13
As they use HCl as an additional acid for the pH adjustment
and electrostatic charging, Renger et al.8,9 utilize TODA only as
a steric dispersant. The advantage of this procedure is that lower
TODA amounts are needed in comparison with the electrosteric
use of it, but the disadvantage is the higher ionic concentration,
especially the presence of Clꢀ ions in the colloid.
There are different processes to achieve separated nanopar-
ticles in colloidal stabilization. The applicability of a process
depends on the original agglomeration state of the particles. A
very efficient route is to use surface modifiers directly in the
particle synthesis process, e.g. to control precipitation nucleation
and growth.2 As the particle synthesis step cannot be included in
every process chain, powders have to be used as raw material
and a surface modification step beyond synthesis is necessary.
One approach is to surface-modify with solely chemical methods
like refluxing.14 But in the majority of cases nanoparticles need
to be deagglomerated mechanically because particles without
stabilization are agglomerated due to Van der Waals attraction,
and without separation, modified agglomerates would result.
Impeller stirring, ultrasonication, or rotor-stator homogeniza-
tion are soft-deagglomeration methods. Ball milling with milli-
meter-sized grinding media in a planetary mill can be applied to
load agglomerates with intermediate forces. Next to kneaders,
stirred media mills provide the highest mechanical forces for
particle treatment.
I. Introduction
ANOSTRUCTURED and nanosized materials feature unique
properties. Therefore, nanoparticles became useful ‘‘carri-
N
ers of functions’’ in material developments that aim at optimiz-
ing manufacturing processes and material properties. In many
cases, beneficial mechanisms based on the size effects of nano-
particles depend on the deagglomeration state of the particles. It
is obvious that only single nanoparticles or very small agglo-
merates yield the desired size range, while agglomerates of many
nanoparticles behave like larger particles. Ceramics and nano-
composites are two important material classes where innova-
tions can be expected by the utilization of nanoparticles.1–3 The
ceramic-shaping process and the resulting sintering behavior is
dependent on the deagglomeration state, as the achieved green
densities increase with decreasing agglomerate sizes. This has
been demonstrated for zirconia agglomerates in the submicro-
meter size range,4 for example. Among other preparation routes,
nanocomposites (with a ceramic or polymer matrix) can be ob-
tained by adding nanoparticles to their manufacturing batches.
Then, the processing of predominantly deagglomerated nano-
particles is a prerequisite for a homogeneously distributed,
nanosized phase in the final composite matrix.
Colloidal stabilization and processing are the keys for
handling single particles and nanoparticles.1,2,5 In order to avoid
agglomeration and to prevent sedimentation, electrostatic
stabilization and/or steric stabilization has to be established
between the particles in the dispersion medium. A first ap-
proach is given by the DLVO theory, exemplarily described by
V. Hackley—contributing editor
Manuscript No. 24413. Received March 13, 2008; approved May 30, 2008.
*Member, The American Ceramic Society.
wAuthor to whom correspondence should be addressed. e-mail: michael.veith@inm-
gmbh.de
2836

September 2008
Milling of Zirconia Nanoparticles
2837
Stirred media milling is known as a suitable ceramic process
method for the comminution (brittle fracture) and the deagglom-
eration of particles with sizes up to the 100-mm range. Achieving
final particle sizes below 1 mm is hindered by reagglomeration if
no special attention is paid to colloidal stabilization.15 Final par-
ticle sizes below 100 nm are achievable by using grinding media
in the size range of 100 mm to 1 mm and using electrostatic sta-
bilization15–17 or by parallel surface modification with electrost-
erically stabilizing molecules.3,11,12 Wear of the grinding media15
as well as wear of the grinding chamber and the stirrer discs is a
side effect in stirred media mills. Further effects to be considered
are changes in the phase composition of the milled material due
to mechanochemical action. Schwedes and Peukert, for example,
observed the formation of aluminum hydroxide during milling of
corundum.16 But the formation of hydroxides will not necessarily
occur for all materials, as these authors did not observe mec-
hanochemical degeneration of crystalline SnO2 particles milled to
sizes in the smaller nanometer range.17
(3) ‘‘TZ’’ is a commercial product (TZ-3Y, Tosoh, Japan)
with 3 mol% Y2O3 doping, produced by hydrolysis of hydrous
zirconium oxychloride and yttrium chloride and subsequent
calcination. Its specified surface area is 1673 m2/g by Tosoh,
and confirmed by our own measurement (13.7 m2/g). As the
powder is almost free from ionic impurities on the surface (con-
ductivity: 90 mS/cm at 5 wt% in water), a washing step was not
executed.
(2) Milling
Milling experiments have been performed in a stirred media mill
(Perl Mill PML-H/V, Drais, Germany, now Buehler Group,
Uzwil, Switzerland). The water-cooled grinding chamber with a
volume of B1 L and the agitator are lined with ZrO2. The sieve
cartridge and the stirrer discs are also consisting of ZrO2. The
mill has been filled with 1.75 kg (B70 vol% filling of the grind-
ing chamber) of ZrSiO4-milling media with a diameter of 0.3–
0.4 mm.
For bulk zirconia, the thermodynamical equilibrium tem-
perature between the tetragonal and the low temperature
monoclinic phase is B11751C. Doping is an important param-
eter to influence phase stability. Besides doping, the resulting
crystalline phase is influenced by particle size. The smaller the
particles, the higher the amount of the tetragonal phase, even in
cases of undoped zirconia. According to the papers,18,19 there
are different possible underlying mechanisms, like the higher
surface energy of the monoclinic phase, internal strain, and ex-
ternal pressure. The surface energy may be affected by adsorbed
water or anions19 and by the surface defect structure. Further, it
seems probable that external stresses, as generated during mill-
ing, may influence the surface structure, the strain, or may have
an effect like isostatic pressure. Experiments with an agate mill20
or with centrifugal planetary ball mills21,22 on ZrO2 nanoparti-
cles and on amorphous zirconium hydroxide are described in the
literature. Milling induced the tetragonal-monoclinic transfor-
mation20 and the reverse reaction22 as well as decreased the
crystallization enthalpy at thermal treatment of amorphous
gels,21 the partial crystallization,21 the crystallization of amor-
phous zirconium hydroxide,22 and the amorphization of the
crystalline phases at prolonged milling treatments.22 These par-
tially contradictory effects depend on the milling parameters
(e.g. dry or in dispersion) and on the previous histories of the
materials. Therefore, regarding these complex dependencies,
here it just shall be concluded that changes in the phase com-
position are also a point to pay attention to during the treatment
of ZrO2 colloids in the stirred media mill.
Two kilograms of the respective ZrO2 powders have been
given to 3 kg H2O with 3,6,9-trioxadecanoic acid (TODA,
Clariant, Sulzbach am Taunus, Germany) as the surface mod-
ifier in a vessel and have been stirred with a magnetic stir bar for
B2.5 days. Milling was executed for 32 h at 3000 stirrer revo-
lutions per minute by continuously circulating the dispersion
through mill and vessel with a membrane pump. The dispersion
in the vessel has been stirred with a magnetic stir bar. The de-
ployed TODA amounts depended on the powder: 300 g for IZ,
157 g for DZ, and 40 g for TZ. Results from the TZ-milling
experiment with a constant TODA amount are labeled with
‘‘TZTODA5 c’’. A second experiment with TZ was started with
the same TODA amount. Then further TODA was added dur-
ing milling to keep the pH constantly at B3.570.1 (indication:
‘‘TZpH5 c’’).
(3) Sample Preparation
Samples (B80–100 mL) have been taken from the dispersions
after different milling times. The ‘‘0 h’’ samples have been
passed through the mill within the first minute of the experi-
ment. Dispersion samples without dilution have been used for
the measurement of the pH, the electrical conductivity, and the
viscosity.
After 32 h of milling the dispersion samples have been ultra-
centrifuged (20 h at 150 000 ꢁ g) to separate coarser zirconia
particles (d42.5 nm) from the supernatant. To check the exis-
tence of molecular (or very fine particulate) Zr, Y, and Si species,
the supernatant has been analyzed using ICP-AES to obtain the
Zr, Y, and Si contents. Samples from the supernatants have also
been put and dried on grids for transmission electron microscopy
(TEM) examination.
Further, dispersions have been diluted for TEM sample prep-
aration, for ICP-AES analysis of the total Si content, and for
particle size analysis using an X-ray disk centrifuge (XDC). To
maintain the colloidal stabilization during the particle size mea-
surements, pH B3 (TODA) water was used for the dilution.
X-ray diffraction (XRD) samples have been prepared by drying
the dispersions. To avoid a disturbance of TODA in dried sam-
ples on the measurement of the specific surface area and on
DSC/TG signals, those samples have been washed according the
following procedure.
For the present work different types of zirconia nanoparticles
have been treated in a stirred media mill. TODA as an electrost-
erically acting surface modifier molecule has been selected for the
colloidal stabilization. It is our aim to describe for the first time,
systematically, the obtainable colloid properties and the effects
caused by the milling process in this material system.
II. Experimental Procedure
(1) Materials
Three different ZrO2 powders have been selected for the exper-
iments.
(1) The ‘‘IZ’’ is an experimental product of our laboratory,
doped with 3 mol% Y2O3 and produced by precipitation of
zirconium propylate and hydrous yttrium nitrate followed by
subsequent hydrothermal crystallization according to pat.12
Afterwards it was washed to reach a conductivity of B10 mS/cm.
IZ particles exhibited a specific surface area of 129 m2/g
(‘‘0 h’’ sample).
(2) Another zirconia type is termed ‘‘DZ’’ in this work. It is
an undoped experimental product (ZrO2 VP, Evonik Degussa,
Frankfurt, Germany). It is produced using flame pyrolysis of zir-
conium tetrachloride. Degussa supplied this powder a few years
ago with a declared specific surface area of 45715 m2/g, which
was confirmed as 45.5 m2/g by our own measurement. Before
milling the powder was washed to a conductivity of B10 mS/cm.
(1) A mixture of 10 g dispersion and 100 g aqueous NaOH
(8 m) was refluxed for 2 h at 1001C.
(2) Afterwards it was decanted, filled with H2O, stirred, and
centrifuged.
(3) The previous step was iterated (four to five times) until a
conductivity of B50 mS/cm was reached, and then the particles
were dried at 1001C.
(4) Analytical Methods
(1) Viscosity measurement (Rheometer MCR 300, Physica,
Stuttgart, Germany) in the cylinder ‘‘DG 42’’ at 201C and at 113
sꢀ1 after applying a shear rate of 1000 sꢀ1
.

2838
Journal of the American Ceramic Society—Adam et al.
Vol. 91, No. 9
Table I. Colloidal Properties Before (Respectively, After 1 h) and After Milling of the ZrO2 Dispersions
TODA amount deployed,
in relation to particle
Milling
time (h)
Viscosity at
113 sꢀ1 (mPaꢂ s)
Conductivity
Milling
time (h)
Mass (wt.-%)
SA (mg/m2)
pH
(mS/cm) (SDo0.02)
d10 (nm)
d50 (nm)
d90 (nm)
IZ
0
32
0
32
0
32
0
32
45
15
1.16
1.15
1.71
1.25
1.43
0.29
1.43
1.71
2.9
3.9
2.5
3.7
3.6
6.3
3.4
3.4
1.5
1.8
1.3
0.8
1.6
0.6
1.6
2.2
1
32
1
32
1
32
1
32
11
19
9
41
24
106
48
111
51
72
13
74
46
174
80
188
89
4.3
4.5
1.9
1.5
2.8
1.9
2.3
d25 5 6w
25
12
DZ
7.85
2
TZTODA5 c
TZpH5 c
69
16
72
17
2
12
wd10 is out of effective measurement range. SA, surface area; TODA, 3,6,9-trioxadecanoic acid.
(2) Si, Zr, Y concentration by ICP-AES (Ultima 2, Horiba
Jobin Yvon, Edison, NJ).
TZ dispersions formed low sediments (o1 mm) after more than
1 month.
(3) TEM imaging (CM200FEG, Philips, Eindhoven, the
Netherlands).
(4) Volume distribution of the colloidal particle sizes by
X-ray disc centrifuge (XDC) (BI-XDC, Brookhaven Instru-
ments, Holtsville, NY).
(5) XRD (D500, Siemens, Munich, Germany), Rietveld re-
finement, and calculations were executed by software (Topas 3,
Bruker AXS, Madison, WI). Primary crystallite sizes result from
the Rietveld fit for 151r2yr901, without considering microstrain.
(6) N2 adsorption for BET-specific surface area determina-
tion (Autosorb-AS6, Quantachrome, Boynton Beach, FL).
(7) Differential scanning calorimetry (DSC) and thermal
gravimetry (TG) (STA 449 C, Netzsch, Selb, Germany) at a
10 K/min heating rate, in air.
Table I lists all the colloidal properties of the dispersions.
The relatively low viscosities after milling can be explained by
their good deagglomeration states and their relatively low solid
contents. The IZ particles lead to the highest viscosity, with
4.3 mPa ꢂ s after milling, due to their high specific surface area.
In the first three experiments, the amount of surface modifier
TODA was kept constant from the beginning. The aim was to
realize electrosteric stabilization at a pH of B4. On first sight,
the TODA amounts in wt% of the particles differ strongly.
But the relevant parameter for comparison is the TODA
amount relative to the particle surface area after milling. We
used our experience that B1 mg/m2 lead to good stabilization,
and Knoll13 observed surface saturation at 1.1 mg/m2 by ad-
sorption measurements of TODA on a similar IZ (former batch)
in ethanol.
An aqueous TODA solution with a concentration as in the IZ
experiment (but without particles) would exhibit a pH slightly
above 1. Before milling, a pH of 2.9 is observed in the batch with
particles. After milling, for IZ, a TODA amount of 1.15 mg/m2
results in a pH of 3.9. One can conclude that during milling
TODA is increasingly bound to the surface, which corresponds
to a decreasing acid concentration in the water. The specific
surface areas of DZ and TZ increased during milling (Table II).
For DZ this was linked with the calculated TODA amount de-
creasing from 1.71 to 1.25 mg/m2 with a pH of 3.7 after milling.
For TZ the newly formed surface area is large, and as a result,
the final pH of 6.3 (at 0.29 mg/m2) is higher, as aimed. There-
fore, in a second TZ experiment, the pH was kept constant by
readjusting with additional TODA during milling. Here, 1.71
mg/m2 TODA has been used, resulting in a pH of 3.4. The
decreasing conductivities during milling of DZ and TZTODA 5 c
are consistent because bound TODA cannot dissociate to ions
contributing to the conductivity. In the case of TZpH5 c, the in-
creased conductivity is based on the high TODA amount. In the
case of IZ, there is a slight increase in conductivity during mill-
ing, which might result from a small fraction of amorphous
material.
The given standard deviations (SD) result from repeated mea-
surements of the same samples, of standards, and from the es-
timation of measurement precision.
III. Results and Discussion
The first part of this chapter is devoted to all the properties of
the resulting colloids, including the achieved particle sizes in the
colloidal state. The second part concerns the comminution effect
of the milling process on the particles, followed by part three,
which describes all other observed effects as a result of milling.
Finally, some information on milling energy is given.
(1) Colloidal Properties and Deagglomeration Effect
Optical appearance of nanoparticular dispersions is a colloid
property, which depends on the state of agglomeration. Before
milling, the dispersions are white and not fully stable against
sedimentation because of the agglomerates. After milling for
32 h, all zirconia dispersions with a solid content of 40 wt%
(10 vol%) are stable, and the appearances have changed in the
cases of IZ and DZ. The IZ dispersion is grayish translucent,
whereas dispersicons based on DZ and TZ are white because
they contain particles 420 nm. Concerning the stability of the
dispersions (B2 L samples), the dispersions based on IZ and DZ
showed no sediments after storage for more than 1 month. Both
The characteristic particle sizes (d90, d50, d10) in the colloids
are decreasing with increasing milling time, as shown in Fig. 1.
The numerical values are given in Table I . The values are
Table II. Crystallite Sizes, Phase Compositions (Tetragonal: t/Monoclinic: m), and Specific Surface Areas (SSA) of ZrO2 Particles
Before and After Milling
Milling
time (h)
dXRD (t) (nm)
(SDo10 %)
dXRD (m) (nm)
(SDo10 %)
t-Fraction (wt.-%)
(SDo0.5)
m-Fraction (wt.-%)
(SDo0.5)
SSA (m2/g)
w
(SDo5 %)
-dSSA (nm)
IZ
0
32
0
32
0
6.7
6.6
16
8.1
19
4.9
5.0
18
14
21
89
82
30
8
71
35
11
18
70
92
29
65
129
130
46
7.8
7.7
22
16
71
DZ
63
TZTODA5 c
14z
70
32
4.8
9.1
14
wCalculated: dSSA (nm) 5 6000/(SSA ꢂ r), with r 5 6 g/cm3, SSA in (m2/g). zMeasurement on powder, as delivered.

September 2008
Milling of Zirconia Nanoparticles
2839
Fig. 1. Particle sizes by X-ray disk centrifuge and pH versus milling time. (For IZ, d25 is displayed instead of d10, which is partially out of the effective
measurement range.)
only given for milling times starting from 1 h because starting
dispersions are too unstable for an exact measurement. All d50
values achieved after a milling times of 32 h are slightly larger
than the average particle sizes that can be calculated from the
specific surface area dSSA (Table II); for example for DZ,
d50 5 24 nm, dSSA 5 16 nm. Therefore, the particles in the dis-
persion are still agglomerates of very few particles, and thus it
can be concluded that the decreasing of colloidal particle sizes is
partly a deagglomeration effect. In the case of IZ, the decrease
of the mean particle size is in fact only based on deagglomer-
ation because there is no increase in specific surface area. This
effect is strong at the beginning of the milling process. Starting
from an agglomerated dispersion with particle sizes partly 41
mm in the first 2 h, the d90 is reduced to o40 nm. In the fol-
lowing 30 h of milling, the d90 is reduced to 13 nm.
In the case of the TZ experiments, the courses of the particle
diameters versus milling time are very similar (Fig. 1, Table I).
Because of the comparatively high final pH of 6.3, which is closer
to the pH of the isoelectric point of Y2O3-doped ZrO2
(pHIEP 56.6–7.2, experimental data from Wei et al.23), it could
be assumed that in the case of TZTODA 5c the dispersion may have
a lower stability. The actual particle size measurement experiments
do not allow conclusions on this assumption, as the dispersions
have been diluted with pH 53 water, acidified with TODA. Only
the low degree of sedimentation for both TZ experiments gives an
indication that the TZTODA 5c dispersion is not less stable.
Figure 2 shows the XDC particle size distributions after
different milling times (1 and 32 h). For all experiments, a clear
shift of the distributions to smaller particle sizes is obvious. In
the case of IZ, the distribution becomes narrower because of
almost complete deagglomeration of the particles. In the case of
DZ and TZpH5 c, the amount of large agglomerates can also be
minimized by milling. Nevertheless, there are still minor frac-
tions of remaining agglomerates after milling for 32 h. Another
observation concerning DZ and TZ is the formation of particle
fractions with diameters o10 nm. These fractions of smaller
particles are most likely a result of comminution.
dispersion medium. As discussed below, wear of the milling
media, particle amorphization, and comminution are potential
causes of these species.
(2) Comminution
Figure 3 shows the XRD patterns of the particles before and
after milling. IZ consists mainly of tetragonal ZrO2 with broad
peaks as a result of small particle sizes, and there are no visible
differences in the diffraction patterns due to the milling process.
DZ consists of monoclinic and tetragonal ZrO2 peaks before
milling, and the tetragonal phase diminishes as a result of mill-
ing. In the case of TZ, the large tetragonal main peak decreases
as well, and a clear broadening of the monoclinic peaks can
be seen.
These effects are confirmed by the data given in Table II. The
particle types represent different mixtures of the tetragonal and
the monoclinic phases. In the experiments described here, mill-
ing causes a partial transformation of the tetragonal to the
monoclinic phase and a decrease of primary crystallite sizes
(XRD) for both phases. The extent of these two effects becomes
larger with increasing starting particle size. The existence of the
tetragonal phase in zirconia nanoparticles is in line with the
generally assumed size effect. If size would be the dominant
parameter, an increase of the tetragonal fraction could be ex-
pected due to comminution. But we observe the opposite effect.
In Fig. 2 it is striking that the cumulative 32 h curves have
values 40% at their beginning. This is due to particles with
diameters o5 nm and/or molecular Zr, Y, and Si species, which
cannot be settled by centrifugation in the XDC. During the
measurement, these fractions remain in the supernatant and
reduce the X-ray transmittance in comparison with the pure
Fig. 2. Particle size distributions by X-ray disc centrifuge of the differ-
ent ZrO2 particle types, after different milling times (1 and 32 h).

2840
Journal of the American Ceramic Society—Adam et al.
Vol. 91, No. 9
SSA values in Table II and leads to the conclusion that zirconia
particles with an average particle size below 10 nm do not un-
dergo comminution due to milling, within the selected process
parameters. After milling, the morphology of DZ particles looks
more square-edged. At higher magnifications, a minor fraction
of very fine, irregular-shaped particles can be observed. How-
ever, based on TEM observation, the particle size distribution of
the coarser particles does not seem to be greatly changed. In
Fig. 4, one larger particle shows two bright stripes, which can be
caused by twinning of the crystallite. Further, larger particles
seem to consist of several crystallites.
After milling, some TZ particles in the TEM appear to be
much smaller than the ‘‘apparent particles.’’ But most of the
‘‘apparent particles’’ maintain their shape, while the inner struc-
ture looks finer, i.e. the ‘‘apparent particles’’ consist of more
crystallites after milling. For DZ and TZ, this results in smaller
crystallite sizes calculated from XRD measurements. The in-
crease in specific surface areas can be explained by a fraction of
smaller individual particles with sizes o10 nm, as can bee seen
in Fig. 2 by the first small peak within the particle size distribu-
tions of DZ and TZ. Their surfaces are accessible for nitrogen
adsorption and thus also contribute to the increased surface
area.
So far, it can be concluded that ZrO2 nanoparticles under-
going the described milling process are deagglomerated, and
comminution effects occur only for the ‘‘coarser’’ DZ and TZ
particles. But comminution does not result in the regular
splitting of all larger particles into two or more particles with
a similar size.
Fig. 3. X-ray diffraction patterns of the particles before and after mill-
ing. The lines pointing upwards (respectively, downwards) show the po-
sitions of the diffraction peaks of the tetragonal (monoclinic) ZrO2
phase.
(3) Milling Side Effects
Apparently, mechanical stress applied on the particles also plays
a role, probably by affecting surface energy (chemistry) or in-
ternal strain.19 In the case of IZ and TZ, the transformation was
possibly supported by a partial leaching of the Y2O3 dopant, as
discussed below.
The decrease of primary crystallite sizes in the case of DZ and
TZ can be ascribed to comminution during the milling process.
Comminution, understood as the splitting of particles, is con-
nected with the creation of new surfaces. Thus, the increasing
specific surface areas listed in Table II confirm comminution for
the coarser DZ and TZ particle types. The primary crystallite
sizes and the specific surface area of the finer IZ particles remain
almost constant.
According to Table II, the mean particle diameters calculated
from the specific surface areas dSSA are larger than the primary
crystallite sizes calculated from XRD measurements dXRD. This
can be explained by the circumstance that the particles consist of
several primary crystallites. In the case of TZ, there are larger
differences between the diameters dXRD and dSSA at the begin-
ning and at the end of the process. In the case of IZ and DZ, the
resulting differences between dSSA (IZ: 7.7 nm; DZ: 16 nm) and
The changes of the particles’ phase composition as discussed in
the previous section can be understood as a milling side effect
next to deagglomeration and comminution.
Another side effect during milling is the wear of milling media.
Amorphous or particulate debris of the ZrSiO4 balls becomes
part of the zirconia nanoparticle dispersions, as it cannot be
separated like the balls themselves. The debris can be quantified
by the Si concentration in the final dispersions. The concentra-
tions and the derived amounts of worn ZrSiO4 balls are given in
Table III (look at columns ‘conc(Si) in dispersion’ and ‘equiva-
lent ZrSiO4 worn’). Up to B36 g of the milling media (2%) are
worn. This amount has to be considered as a potential impurity
in the dispersion, but it is not an unexpected high value, because
of the high mechanical load in the stirred media mill. It is in line
with similar experiments,15 where a wear of up to 10% of the
milling media occurred because of intense milling conditions.
Amorphization of material from the particles is another pos-
sible (material depending) side effect of milling, as discussed in
the literature survey. This consideration, and the proof of a mi-
nor fraction of fine particles and/or molecular compounds
smaller than 2.5 nm by XDC, led to the search for amorphized
material as a result of milling. Possible mechanisms to consider
are the transformation of crystalline into amorphous particles
and the mechanically induced reaction of zirconia to zirconium
hydroxide, which would also occur as an amorphous fraction
(after drying).
Neither the occurrence of ‘‘amorphous humps’’ in the diffrac-
tion patterns (Fig. 3) nor peaks in the DSC curves of milled and
organic-free powders could be detected. The TG curves show a
continuous weight loss of o2% in sum between 1201 and 9001C,
deriving from chemisorbed water. No stepwise weight loss from
dehydration of amorphous zirconium (oxide) hydroxides was
observed. These three results allow the conclusion that no large
particle fraction (e.g., 45%) has been transferred to amorphous
compounds.
In order to gain more information about the minor fine frac-
tion, the larger particles (42.5 nm) were separated from the
dispersions by ultra centrifugation. The TEM solids from
the supernatant mainly appear amorphous with undefined
shapes. Only a small fraction of small crystalline particles can
dXRD (IZ: 6.6 nm/tetragonal; DZ: 14 nm/monoclinic) after mill-
ing for 32 h are relatively small, indicating that these particles
are mostly single crystallites and their distributions are nearly
monomodal.
The larger differences for TZ between dSSA (0 h: 71 nm; 32 h:
14 nm) and dXRD (0 h: 19 nm/tetragonal; 32 h: 9.1 nm/mono-
clinic) can be explained by means of TEM (Fig. 4). Without
milling, individual IZ and DZ particles can be recognized
and they are mainly single crystallites. In the case of DZ,
some sintering necks resulting from the flame pyrolysis process
can be seen. TZ particles have a different appearance because,
here, spherical shapes in the size range of 40–60 nm (termed as
‘‘apparent particles’’ by the manufacturer and in the following)
can be recognized as compact aggregates that are sintered to-
gether. Each of these sintered aggregates consists of several crys-
tallites, so that the specific surface area is determined by the
‘‘apparent particles.’’
The morphology of the IZ particles has not changed during
milling. This is consistent with the nearly unchanged dXRD and

September 2008
Milling of Zirconia Nanoparticles
2841
Fig. 4. Transmission electron miroscopy micrographs of the ZrO2 nanoparticles before (left side) and after (right) milling. TZ particles [(e), (f)] come
from the TZTODA5 c experiment.
be observed, but the TEM does not allow conclusions on the
quantitative relations of the fractions.
Therefore, the concentrations of Zr, Y, and Si in the super-
natant have been measured by ICP-AES. The results are listed in
the left part of Table III.
The largest Zr concentration (9.3 g/L) in the ultra centrifu-
gation supernatant was found for the IZ sample. From this
value it can be roughly calculated that 2% of the Zr atoms from
the zirconia nanoparticles were transferred into a molecular or a
very fine particulate (do2.5 nm) form.

2842
Journal of the American Ceramic Society—Adam et al.
Vol. 91, No. 9
Table III. ICP-AES Examinations on Dispersions, Respectively, on their Supernatants After Ultracentrifugation
and After Milling for 32 h
Measured elemental concentrations in
Calculated from Si concentration in dispersion
Zr concentration
Equivalent ZrSiO4
from milling media
y Supernatant
y Dispersion
worn in full-milling batch
(Si5 Zr, molar)
Zr (g/l)
Y (g/l)
Si (g/l)
Si (g/l)
(g)
(%)
Zr (g/l)
IZ
DZ
9.3 (0.13)
4.4 (0.09)
0.3 (0.01)
5.0 (0.11)
2.8 (0.04)
1.1 (0.03)
1.6 (0.03)
1.5 (0.03)
0.9 (0.01)
0.7 (0.01)
36
34
19
15
2.0
1.9
1.1
0.9
5.3
5.0
2.9
2.3
B0
o0.1
3.7 (0.16)
o0.1
TZTODA5 c
TZpH5 c
B0
o0.1
The wear of ZrSiO4 milling balls also contributes to the Zr
concentration in the supernatant if the debris occurs in a mo-
lecular or particulate (do2.5 nm) form. The maximum possible
Zr concentration resulting from the debris can be calculated
from the Si concentration in the dispersion and is given in the
last column of Table III.
From these considerations it can be concluded that only small
fractions (fa) of the zirconia nanoparticles have been trans-
formed into amorphous compounds during milling (for IZ: fa
o2.0% if only the nanoparticles contribute to the Zr amount in
the supernatant, and fao0.7% if the wear of milling media is
taken into account). For the DZ and TZ samples, the Zr con-
centrations in the supernatant and the possible amorphous frac-
tions are even smaller. Such small fractions are consistent with
the result that no traces of amorphous material were found by
XRD and DSC/TG.
Table III also lists the measured Si concentrations in the su-
pernatant. They are less than the Si concentrations in the dis-
persion. This can be explained by a partial inclusion of the Si in
the sediment, for example, as a result of a partial adsorption of
Si compounds on zirconia nanoparticles.
Acidic leaching of yttrium from the particles can be concluded
from the data on Y concentrations in the supernatant (Table III)
and thus constitutes another milling side effect. For IZ, a frac-
tion of 10% of the Y atoms (zirconia doping) can be found in
the dispersion medium in a molecular or particulate (do2.5 nm)
form. As discussed above, this applies only to 2% of the Zr
atoms. Thus, the excess Y amount may be explained by acidic
leaching of Y from the particles during milling. This is con-
sistent with the occurrence of the highest Y concentration in
the case of TZpH 5 c (pHB3.5) and the lowest Y concentration
(for the doped types) in the case of TZTODA 5 c (pH 5 6.3 at
the end).
IV. Conclusions
The processing of zirconia nanoparticles in stirred media mills
with an in situ surface modification is a very effective method for
the production of stable colloids with almost completely deag-
glomerated particles.
Applying a stirred media milling process on nanoparticles, the
question arises if comminution is possible. For a given milling
media size (in this work: 0.3–0.4 mm), a lower particle size
threshold seems to exist. In contrast to the case of the smaller IZ
particles, the creation of smaller fragments and crystallites of
DZ and TZ can be observed, but larger primary particles (single
crystallites) or ‘‘apparent particles’’ (hard aggregates) are mostly
not comminuted, as it is usually affected by milling of much
larger particles. The reason, therefore, is the lower probability of
hitting smaller particles with milling media balls of a given size
and the relatively low impact of the lightweight milling balls.
Concerning the observed side effects, it depends on the par-
ticular application of nanoparticles, if variances in the crystalline
phase composition, impurities due to wear of milling media (and
milling chamber), and a small fraction of amorphous zirconium
(oxide) hydroxide can be accepted. A contamination with silicon
due to ZrSiO4 balls can be avoided by using commercially avail-
able zirconia milling media. The leaching of yttrium can be
minimized by minimizing the TODA amount not exceeding the
particle surface coverage, i.e. at less acidic conditions.
Processing of nanoparticles with stirred media mills is a com-
plex process that depends on geometrical parameters (sizes of
particles and milling media), physical parameters (energy input,
solid content of the dispersion, deployed milling media amount),
and chemical properties (particle composition, surface modifier,
pH, dispersion media).
Altogether, an interesting colloidal-processing method is pre-
sented. It offers the opportunity for further research, for exam-
ple, concerning other particle or surface modifier types and the
optimization of the surface modifier amount.
(4) Applied Energy in the Milling Process After 32 h
In the experiments with a stirred media mill described by Mende
and colleagues,15–17 specific (with regard to the particle amount)
milling energies up to 60 MJ/kg have been applied. This energy
amount was derived from the data of a torque sensor on the
stirrer shaft, and hence it can be understood as mechanical en-
ergy information. The mill used for the present paper is not
equipped with a torque sensor, but it displays the electrical
effective power consumption. A specific electrical milling energy
of 6 MJ/kg can be calculated for the experiments with TZ. The
mechanical energy in this work cannot be calculated due to
missing efficiency factors of engine and transmission, but it must
be smaller than 6 MJ/kg. In comparison with Mende and col-
leagues,15–17 o10% of their milling energy was applied. This is
consistent with Mende and colleagues,15–17 having the focus on
the comminution of coarser alumina (micrometer sized) as the
starting material and using coarser milling media. In contrast, in
the present work, mainly a deagglomeration effect has been
achieved.
Acknowledgments
We especially thank J. Dietz for perfectly performing the milling experiments,
much of the sample preparation steps, and part of the analytics. We greatly ap-
preciate the analytical work and results contributed by M. Koch (with special
thanks for the comprehensive discussion of TEM micrographs), A. Jung, C. Fink-
Straube, and I. Grobelsek.
References
1H. K. Schmidt, ‘‘Relevance of Sol–Gel Methods for Synthesis of Fine Parti-
cles,’’ KONA Powder Particle, 14, 92–103 (1996).
2H. K. Schmidt, ‘‘Nanoparticles by Chemical Synthesis, Processing to
Materials and Innovative Applications,’’ Appl. Organomet. Chem., 15, 331–43
(2001).
3H. K. Schmidt, F. Tabellion, K.-P. Schmitt, and P.-W. Oliveira, ‘‘Ceramic
Nanoparticle Technologies for Ceramics and Composites,’’ Ceram. Trans., 148,
173–86 (2004).
4D.-M. Liu and J.-T. Lin, ‘‘Influence of Ceramic Powders of Different Char-
acteristics on Particle Packing Structure and Sintering Behaviour,’’ J. Mater. Sci,
34, 1959–72 (1999).

September 2008
Milling of Zirconia Nanoparticles
2843
5A. R. Studart, E. Amstad, and L. J. Gauckler, ‘‘Colloidal Stabilization
of Nanoparticles in Concentrated Suspensions,’’ Langmuir, 23 [3] 1081–90
(2007).
14R. Nass, H. K. Schmidt, and K.-P. Schmitt, ‘‘Method of Manufacturing
Surface-Modified Ceramic Powders with Particles in the Nanometre Size,’’ Euro-
pean Patent EP 0636111 B1.
6R. C. D. Cruz, J. Reinshagen, R. Oberacker, A. M. Segada
˜
es, and M. J.
15S. Mende, F. Stenger, W. Peukert, and J. Schwedes, ‘‘Mechanical Production
and Stabilization of Submicron Particles in Stirred Media Mills,’’ Powder Technol.,
132, 64–73 (2003).
Hoffmann, ‘‘Electrical Conductivity and Stability of Concentrated Aqueous
Alumina Suspensions,’’ J. Colloid Interface Sci., 286, 579–88 (2005).
7H. Y. T. Chen, W. C. J. Wei, K. C. Hsu, and C. S. Chen, ‘‘Adsorption of PAA
on the a-Al2O3 Surface,’’ J. Am. Ceram. Soc., 90 [6] 1709–16 (2007).
8C. Renger, P. Kuschel, A. Kristoffersson, B. Clauss, W. Oppermann, and
W. Sigmund, ‘‘Rheology Studies on Highly Filled Nano-Zirconia Suspensions,’’
J. Eur. Ceram. Soc., 27, 2361–7 (2007).
16F. Stenger, S. Mende, J. Schwedes, and W. Peukert, ‘‘The Influence of Sus-
pension Properties on the Grinding Behaviour of Alumina Particles in the Sub-
micron Size Range in Stirred Media Mills,’’ Powder Technol., 156, 103–10 (2005).
17F. Stenger, S. Mende, J. Schwedes, and W. Peukert, ‘‘Nanomilling in Stirred
Media Mills,’’ Chem. Eng. Sci., 60, 4557–65 (2005).
9C. Renger, P. Kuschel, A. Kristoffersson, B. Clauss, W. Oppermann, and
W. Sigmund, ‘‘Adsorption Studies on Nano-Zirconia in Water and a Water-1,2-
Propanediol Mixture,’’ J. Ceram. Process. Res., 7 [2] 106–12 (2006).
10A. R. Studart, E. Amstad, M. Antoni, and L. J. Gauckler, ‘‘Rheology of
Concentrated Suspensions Containing Weakly Attractive Alumina Nanoparti-
cles,’’ J. Am. Ceram. Soc., 89 [8] 2418–25 (2006).
18T. Chraska, A. H. King, and C. C. Berndt, ‘‘On the Size-Dependent Phase
Transformation in Nanoparticulate Zirconia,’’ Mater. Sci. Eng., A286, 169–78 (2000).
19S. Shukla and S. Seal, ‘‘Mechanisms of Room Temperature Metastable
Tetragonal Phase Stabilization in Zirconia,’’ Int. Mater. Rev., 50 [1] 45–64 (2005).
20T. Mitsuhashi, M. Ichihara, and U. Tatsuke, ‘‘Characterisation and Stabili-
zation of Metastable Tetragonal ZrO2,’’ J. Am. Ceram. Soc., 57 [2] 97–101 (1974).
11J. Adam, K. Gossmann, H. K. Schmidt, K.-P. Schmitt, and F. Tabellion,
‘‘Chemomechanical Production of Functional Colloids,’’ Patent Application EP
1592503 A1.
21J. M. Criado, M. Gonza
J. Malek, ‘‘Influence of Environment and Grinding on the Crystallisation Mech-
anism of ZrO2 Gel,’’ J. Phys. Chem. Solids., 68, 824–9 (2007).
lez, M. J. Dianez, L. A. Perez-Maqueda, and
´ ´ ´
´
12H. K. Schmidt, K.-P. Schmitt, K. Schmitt, and F. Tabellion, ‘‘Method for
Production of Nanoparticles with Custom Surface Chemistry and Corresponding
Colloids,’’ Patent Application EP 1797006 A2.
22P. N. Kuznetsov, L. I. Kuznetsova, A. M. Zhyzhaev, G. L. Pashkov, and V. V.
Boldyrev, ‘‘Ultra Fast Synthesis of Metastable Tetragonal Zirconia by Means of
Mechanochemical Activation,’’ Appl. Catal. A: Gen., 227, 299–307 (2002).
23W.-C. J. Wei, S.-C. Wang, and F.-Y. Ho, ‘‘Electrokinetic Properties of Col-
loidal Zirconia Powders in Aqueous Suspension,’’ J. Am. Ceram. Soc., 82 [12]
13S. C. M. Knoll, ‘‘(Electrophoretic shaping of nanoscaled zirconia) El-
ektrophoretische Formgebung von nanoskaligem Zirkoniumdioxid,’’ PhD Study,
Saarland University, 2001.
3385–92 (1999).
&